Range Profiler readability suggestion


First, congratulation for the nice work on the Range Profiler in 5.4. It is going in the right direction for sure! At least for the usage we intend to get with it.

I would like to point you out a very anoying issue regarding the readalibity of the different profile category. It is in fact very useful to get a compare view of, for instance, Range Programs and Perf Markers at the same time, which is very easy to do with your UI.

However, depending of the renderer, we can end-up with a huge vertical stacking of Markers or Programs that completely dismissed the ability to use the UI correctly (with the split windows for comparing results in same UI). We can clone window to get a compare, but it is less responsive and become a bit messy with tiny and packed stuff to read. Also, zooming the timeline won’t work synchronously for both of them.

It would be awesome if resizing the categories don’t resize the Marker’s bar so it will continue to display names and provide ability to set the UI in “compare mode”. Actually, when the category has not been scaled like crazy (at least on our end), the different icons will start to be too tiny and won’t display text anymore. In addition to that, when we are in a big capture (lots of draws), the UI response is quite affected so scaling or zooming is starving the workflow. If we can set a nice UI layout, it will be done once and then afterwards get a better access to the tool.

Here is an screenshot that shows the issue on our end.


Keep up the good work! We can feel in this version that it begins to be solid with a clearer view of the tool mindset.


Thanks for your feedback.
We’ll take a track for your request and let you know if there are some update.



Hi developers! 'm bumping this suggestion in the hope that this will be tackled in the next release.

I would also ask you a question.

Do you have any plan of reworking the range profiller view in a way it’ll become more flexible to display contention and parallelism? In our engine the readability is totally off so we’re not using NSight for the PC version of our product. Also, the way our engine is made generates way more drawcalls that in a more common one (like Unreal or Unity). Performance are so bad that capturing a simple frame takes more then a hour.

Do you have any cue that you would be able to share with us? I’m met your team at GDC 2015. I’ve exposed our issues and the upcoming development was supposed to tackle these kind of topics. Unfortunately since that time NSight remain unusable for our AAA project.

In the hope of introducing NSight in our pipeline in a near future!